agentleFS
samulee003samulee003/agentic-ppt-skills/.agents/skills/apply-comments/SKILL.md

apply-comments

Skill0 starsChanged 25 days agoMITView on GitHub

Apply pending @slide-comment markers written by the open-slide inspector tool. Use when the user asks to "apply comments", "process slide comments", "apply the inspector comments", or references markers left inside `slides/<id>/index.tsx`.

The file

Show its rules and text

We split this skill into the rules it gives an agent

13 rules. About 0 are common, 5 are distinctive, and 8 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 8

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Always sits on its own line as the first child inside the JSX element it refers to (i.e.

    line 20
  2. Very specific:

    If the user names one (getting-started, q2-roadmap, etc.), work on that single slides/<slideId>/index.tsx.

    line 31
  3. Very specific:

    If they say "all" or don't specify, scan every slides/*/index.tsx.

    line 32
  4. Distinctive:

    Run the regex above against the whole file.

    line 35
  5. Distinctive:

    If there are no markers, tell the user and stop.

    line 38
  6. Very specific:

    Read enough surrounding code (parent element, sibling elements, inline styles) to apply the change faithfully.

    line 42
  7. Very specific:

    If the note is ambiguous, do the smallest reasonable interpretation and mention the assumption in your summary.

    line 43
  8. Very specific:

    Delete the entire marker line including its trailing \n.

    line 50
  9. Distinctive:

    Never leave a marker behind for an edit you applied — that signals a failure.

    line 51
  10. Distinctive:

    After all edits, re-read the file and confirm the only remaining markers are ones you reported as skipped.

    line 54
  11. Very specific:

    Do not touch package.json, open-slide.config.ts, or files outside slides/.

    line 80
  12. Distinctive:

    Do not add dependencies.

    line 81
  13. Very specific:

    Do not re-introduce markers or leave TODO breadcrumbs — the user already has a record in git.

    line 82

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Apply slide comments

The open-slide editor has an inspector tool that lets the user click on a rendered page element and attach a textual comment (e.g. "make this red", "change to 'Open Slide Rocks'"). Each comment is persisted as an in-source JSX marker inside slides/<slideId>/index.tsx.

Your job: read those markers, perform the described edits, and delete the markers.

Before making any page edit, consult the slide-authoring skill — it is the technical reference for how slides/<id>/index.tsx is structured (canvas, type scale, palette, assets, file contract). A comment like "make this bigger" or "change the accent colour" should be applied in a way that stays consistent with those rules.

Marker format

{/* @slide-comment id="c-<8hex>" ts="<ISO>" text="<base64url(JSON)>" */}
  • Always sits on its own line as the first child inside the JSX element it refers to (i.e. between that element's opening > and its other children). The marker is dropped into its target, not floated above it.
  • text is base64url-encoded JSON: {"note": "...", "hint"?: "..."}.
  • Detection regex (authoritative — use exactly this):

    /\{\/\*\s*@slide-comment\s+id="(c-[a-f0-9]+)"\s+ts="([^"]+)"\s+text="([A-Za-z0-9_\-]+={0,2})"\s*\*\/\}/g

Procedure

  1. Identify the target slide(s).
    • If the user names one (getting-started, q2-roadmap, etc.), work on that single slides/<slideId>/index.tsx.
    • If they say "all" or don't specify, scan every slides/*/index.tsx. Process each slide one at a time.
  2. Read the file and find all markers.
    • Run the regex above against the whole file.
    • For each match, base64url-decode text and JSON.parse it to get { note, hint? }.
    • Record each hit as { id, lineIndex (0-based), note, hint }.
    • If there are no markers, tell the user and stop.
  3. Understand each comment in context.
    • The targeted JSX element is the enclosing element of the marker — i.e. read upward from the marker line until you reach the unclosed JSX opening tag whose body the marker lives in. That element is the target. (For self-closing elements like <img />, the inspector hoists the marker to the nearest non-self-closing ancestor; in that case the comment usually refers to a child of the enclosing element rather than the enclosing element itself — use the note text to disambiguate.)
    • Read enough surrounding code (parent element, sibling elements, inline styles) to apply the change faithfully. A comment inside a <div> with an inline background style usually refers to that element's styling, for example.
    • If the note is ambiguous, do the smallest reasonable interpretation and mention the assumption in your summary.
  4. Apply edits in reverse line order.
    • Sort markers by descending lineIndex and process one at a time, using the Edit tool.
    • Processing top-down would invalidate line numbers for later markers as the file shrinks/grows.
  5. Remove each marker after applying its edit.
    • Delete the entire marker line including its trailing \n.
    • Never leave a marker behind for an edit you applied — that signals a failure. Markers deliberately skipped per the edge cases below stay in place.
  6. Verify.
    • After all edits, re-read the file and confirm the only remaining markers are ones you reported as skipped.
    • Confirm the edited JSX is well-formed (balanced tags, no dangling attributes). If the project's package.json has typecheck/lint scripts, run them with the project's package manager; scaffolded projects ship neither TypeScript nor a linter — there, rely on the running dev server (or the build script) to surface compile errors. Fix any errors you introduced.
  7. Report.
    • Summarise: N applied, M skipped plus a one-line description of each change (including the slide id).

base64url decoding helper

function decode(s) {
  const pad = s.length % 4 === 0 ? '' : '='.repeat(4 - (s.length % 4));
  return Buffer.from(s.replace(/-/g, '+').replace(/_/g, '/') + pad, 'base64').toString('utf8');
}

You can run this inline via node -e '...' if you need to inspect a payload; otherwise just reason about the decoded string.

Edge cases

  • Marker with no enclosing JSX element (shouldn't happen — the inspector won't write one — but if you find one): delete it and note as orphan.
  • Multiple markers stacked on consecutive lines inside the same element: they all refer to that enclosing element. Read their notes in source order to understand the combined intent, then apply and delete them bottom-up per step 4.
  • Comment asks for something outside the target element's scope (e.g. "add a new page"): do the closest-reasonable edit and mention the scope expansion in your summary.
  • Can't resolve the comment (e.g. truly ambiguous, or the file changed shape such that the target element doesn't exist): leave the marker in place and report it as skipped. Don't guess.

Do not

  • Do not touch package.json, open-slide.config.ts, or files outside slides/.
  • Do not add dependencies.
  • Do not re-introduce markers or leave TODO breadcrumbs — the user already has a record in git.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.